I'm running some tests to determine what the best approach is as far as number of queues is concerned (see other topic - number of queues and topics).
One thing that I've discovered is that after creating a number of topics (done via the JMX core server API), the CPU utilization of the hornetq server process progressively rises - at 10,000 topics (with no subscribers at this point), the CPU utilization I'm seeing is a constant 25% or so. As I scale it up, my CPU utilization increases to the point that at about 50,000 queues I'm running at 95-100% CPU utilization even though absolutely no traffic is going through the server.
Is this something that is expected, or is this concerning?
Oh, I should mention that I'm running this in a clustered configuration - 2 nodes are active.